Benchmark
Sayf CyberMetric-500 accuracy
Higher scores are better.
Benchmark sourceScores from Sayf cybersecurity evaluations data docs.
- Models measured
- 6
- Scores recorded
- 6
- Sources
- 1
- IQ weight
- 1.64
Score distribution
- 0.947
- Mean
- 0.956
- Median
- 0.016
- Std dev
- 0.93
- 10th percentile
- 0.962
- 90th percentile
- 0.92
- Min
This benchmark has a weight of 1.64 for chat models, measured over 6 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6cybermetric· Sayf cybersecurity evaluations data docs | 0.962 |
| 2 | GPT 5.4cybermetric· Sayf cybersecurity evaluations data docs | 0.96 |
| 3 | Qwen3.6 35B A3Bcybermetric· Sayf cybersecurity evaluations data docs | 0.956 |
| 4 | Gemma 4 31B ITcybermetric· Sayf cybersecurity evaluations data docs | 0.952 |
| 5 | Llama 3.3 70B Instructcybermetric· Sayf cybersecurity evaluations data docs | 0.93 |
| 6 | GPT OSS 20bcybermetric· Sayf cybersecurity evaluations data docs | 0.92 |
Data credits: Sayf-eval. Copyright (c) 2026 Qatar Computing Research Institute (QCRI). MIT License. (licence)